What if the model is no longer your competitive edge and the real moat is the context only your institution owns?

In this episode of Re(AI)magine Conversations, Barath Narayanan, Global BFSI and Europe Geo Head at Persistent, speaks with Ray Wang, Founder, Chairman and Principal Analyst at Constellation Research, about why 2026 marks the shift from AI promise to execution in financial services.

Across banking, insurance, capital markets and payments, the focus is moving beyond experimentation. Financial institutions are looking for measurable improvements in decision speed, operational efficiency, regulatory compliance and cost. Ray shares that BFSI leaders are seeing AI-driven productivity gains of 10X, 100X and, in some cases, 1,000X.

The opportunity extends beyond doing the same work faster. AI is compressing decision cycles across counterparty risk, hedging, credit, regulatory compliance, trading and market identification. Decisions that once took days can increasingly be made in seconds or milliseconds. This changes how financial institutions compete and how quickly their people must respond.

As frontier models become increasingly commoditized, the source of differentiation is also changing. Expertise is becoming widely accessible. Experience and enterprise context are not. Many organizations have invested heavily in models and individual agents while investing less in the institutional knowledge needed to connect them.

For a bank or insurer, that knowledge includes underwriting rules, claims history, regulatory posture, business processes, decision logic and the relationships embedded across its data and systems. This proprietary context enables AI to understand how the institution operates and produce outcomes that are relevant, consistent and defensible.

Barath explains Persistent’s framework of Core, Context and Coordination for building this architecture.

• Core provides the secure and governed AI foundation. It connects data and workflows while supporting model routing, observability, guardrails, token economics and cost control.
• Context gives agents reliable and traceable access to the institution’s own data, processes, rules and history. It provides the domain awareness needed to turn institutional knowledge into an enterprise asset.
• Coordination brings people, agents and systems together within governed, auditable and process-driven workflows. It enables them to use shared context and work towards a consistent outcome.

The need for shared context becomes more important as institutions deploy multiple agents. When agents approach the same task using different information or decision logic, they can produce inconsistent results. In financial services, these outcomes must be repeatable, explainable and defensible, particularly when they affect credit decisions, claims, fraud actions or regulatory obligations.

The episode brings this architecture to life through real-world examples. Barath discusses a multi-agent insurance claims flow spanning first notice of loss, fraud detection and estimation. With each agent working from a shared digital representation of the adjudicator, the claims cycle was reduced from five to seven days to under ten minutes while helping decisions remain consistent across the process.

He also describes a payments modernization initiative where Persistent was brought in to establish a context layer for a specific micro-domain. The approach was estimated to reduce the required effort to roughly one-quarter of the original projection.

The discussion reinforces why context should be built one micro-domain at a time. Rather than starting with hundreds of disconnected use cases, institutions can select one end-to-end business outcome, establish the required Core and build the Context before adding more agents. Each domain then creates a foundation that can support further use cases.

Ray and Barath also examine how enterprise AI is changing commercial models. Customer expectations are moving away from talent-based engagements towards end-to-end outcomes. Bespoke solutions, outcomes-based pricing and joint ventures are becoming more relevant as organizations look for partners that can understand their existing estate and take responsibility for solving complex business problems.

The shift is from running AI at scale to running it coherently at scale. Scale is a volume problem. Coherence is an architecture problem. The institutions that create shared context and coordinate agents, people and systems as one enterprise will build a stronger and more durable advantage.

Tune in to learn why context is becoming BFSI’s real AI moat and how Persistent helps financial institutions build trusted, coordinated and business-aware AI one micro-domain at a time.

Join the conversation. Contact us at podcasts@persistent.com.

Speakers 

Barath Narayanan, Global BFSI and Europe Geo Head, Persistent

Ray Wang, Founder, Chairman and Principal Analyst, Constellation Research